libri-alpha-0.75-Temp-1-attention-3-layers-distil-with-6-layers-take-3
This model is a fine-tuned version of rohitp1/libri-alpha-0.75-Temp-1-attention-3-layers-distil-with-6-layers-take-2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 236.1198
- Wer: 0.2607
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.3
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 388.1305 | 0.45 | 200 | 228.0258 | 0.2599 |
| 376.7096 | 0.9 | 400 | 226.8922 | 0.2566 |
| 384.1615 | 1.35 | 600 | 228.5904 | 0.2571 |
| 373.8909 | 1.79 | 800 | 229.0286 | 0.2563 |
| 385.2149 | 2.24 | 1000 | 230.8802 | 0.2575 |
| 384.5473 | 2.69 | 1200 | 230.1264 | 0.2563 |
| 383.9426 | 3.14 | 1400 | 232.5964 | 0.2569 |
| 385.9253 | 3.59 | 1600 | 237.4036 | 0.2599 |
| 396.9868 | 4.04 | 1800 | 236.1198 | 0.2607 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1
- Datasets 2.6.1
- Tokenizers 0.13.1
- Downloads last month
- 7